Is Bisexuality a Political Statement?

The Center for the Study of Partisanship and Ideology (CSPI) recently released statistics showing that between 2010 and 2021, there was a massive surge of young women identifying as bisexual, but only having sex with men. This was immediately enveloped in the culture war.

Bari Weiss (who has been married to both a man and a woman) said, “I don’t care how sensitive your boyfriend is, and how much he loves listening to Ani DiFranco with you. That doesn’t count.”

Scott Alexander argued that these young women really are bisexual, and critiqued people accusing them of faking bisexuality, saying that there is “a great statistical explanation for why it’s tempting to believe this, but why it isn’t true.” This is a controversial topic, and I appreciated that he both found compassion for people identifying as bisexual and also cared about statistically accuracy. Many commentators accomplished neither.

Rather than approach this through an ideological lens, I’d like to just walk through Scott Alexander’s argument, because it highlights just how bizarre these statistics really are.

His argument is that if a bisexual woman chose her partners at random based on a pool of available partners interested in women, then that pool would be roughly 90% male and 10% female (the lesbian and bisexual women interested in dating other women). He says that most people have seven partners before marriage, so he models this as the probability of randomly selecting seven consecutive male partners out of that total dating pool:

0.9 ^ 7 = 0.48

Which is roughly in the ballpark of the observation that 55% have had only male partners and would indicate that these women identifying as bisexual really are attracted to other women, but there are just a lot more men available.

I thought this seemed very reasonable. And if that is the rationale then we should do the same for 2008 - 2010 numbers, which was 13.3% never having a female partner. If we solve for x in:

x ^ 7 = 0.133

we get x = 0.75.

But how do we get 0.75 instead of 0.9? It would indicate that in the time prior to 2008 that 32% of women were identifying as bisexual or lesbian, making their dating pool three quarters male instead of 90% male. It means that female bisexuality was at 32% in 2008 and has dramatically decreased to 10%. Except that the opposite is happening: the share of non-heterosexual women has rapidly increased.

Another problem is that Scott Alexander’s statistics are deeply out of sync with two larger trends happening in the United States. First, colleges have skewed to roughly only 40% male graduates. Second, the surge in young women identifying as LGBT is highly concentrated in small geographic areas, namely liberal universities. The combination of these two trends mean that many bisexual women have a dating pool that is roughly equal male and female, not 90% male.

Identifying as LGBT became highly politicized around 2016:

And liberals are not uniformly distributed across the United States. A typical voting map that just shows a binary blue or red does not capture the extremes. Remember that slightly liberal is at 95% heterosexual, the same as the other political groups. It’s liberal that’s at 66%. So look at the dark blue counties, not the light blue counties.

Additionally, since this is for ages 18 - 29, the average age would be around 24. This would put them just post-college, and men make up only about 41% of college students. I’ve written before about this gender ratio. For a point of reference, after the First World War, the UK had 67 surviving males for every 100 women in the 20 - 29 age range. Soviet Russia had 72 after the Second World War. Right now, college campuses have about 70 men for every 100 women.

You might say, well what about after college? This gender divide is persisting after graduation because some areas have large numbers of college graduates. In his 2015 book, Date-onomics, Jon Birger looked into the impact of the gender ratio of US college graduates on dating and found that “In Manhattan, the pool of 24-and-under college grads has 38 percent more women than men. In Raleigh, North Carolina, the gap is 49 percent. In Miami and Washington, D.C., it’s 86 percent and 49 percent, respectively” (Date-onomics, 6). Note that this extra 38% female in Manhattan isn’t on a college campus, this is on the entire island of Manhattan. And it’s only graduates, not current students. These numbers were from a decade ago and have gotten more extreme since.

Let’s take a liberal college that’s 41% male with liberal women at 34% LGBT. For 100 students at the college, that’s roughly 38 men interested in women. And 20 women interested in dating women. A bisexual woman’s dating pool is now only 66% male. But as we dig deeper into the data we find that this estimation of a 66% male dating pool might actually be too high.

I’m going to ignore the two columns on the far right because they don’t have as much of a sample size. Let’s focus on the third from the right, “No Religion, Female, Very Liberal” with a sample size of 3,642. That’s at 59% LGBT.

Let’s take UCLA as an example. They have almost exactly the national average of 41% male undergrads and has a very liberal student body, so we can safely assume that the female students match the sample above with 59% of the female population as LGBT. Then a woman would have a dating pool of 53% male. And yet, 55% of them have only had male partners.

This means only 1.2% of bisexual women should have randomly selected seven consecutive male partners:

0.53 ^ 7 = 0.012

This 53% estimate is likely not the national average for bisexual women. Not all are in college, or living in blue counties. But the majority of them are.

Scott Alexander lists a number of other reasons why bisexual women might be more inclined to prefer male partners: homophobia, easier making biological children, men are more assertive at asking women out on dates. But those things were all true in 2008 as well and provide no reason why these numbers should change.